support/scripts/genimage.sh: pass an empty rootpath to genimage

genimage makes a full copy of the given rootpath to ${GENIMAGE_TMP}/root
so passing TARGET_DIR would be a waste of time and disk space. We don't
rely on genimage to build the rootfs image, just to insert a pre-built
one in the disk image.

Signed-off-by: Carlos Santos <unixmania@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

configs/engicam_imx6qdl_icore_qt5: Remove the swrast option

i.MX6 has GPU in hardware, so we really don't need to
fallback to software rendering.

Remove the unneeded BR2_PACKAGE_MESA3D_GALLIUM_DRIVER_SWRAST
option.

Signed-off-by: Fabio Estevam <festevam@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

configs/engicam_imx6qdl_icore_qt5: Select the KMSRO gallium driver

After migrating to Mesa 19.2 the imx-drm_dri.so is no longer installed
unless the kmsro gallium driver is selected, so explicitly select
BR2_PACKAGE_MESA3D_GALLIUM_DRIVER_KMSRO to fix the problem.

Signed-off-by: Fabio Estevam <festevam@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/dbus-python: bump to version 1.2.12

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/dbus-python: remove unrecognized options

--disable-html-docs and --disable-api-docs has been removed since
version 1.2.8 and
https://github.com/freedesktop/dbus-python/commit/721ef6373069bfc6955018681204e20e1ad836e4

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/python-scapy3k: remove 'v' prefix

Fixes version parsing for release-monitoring.org support.

Signed-off-by: Matthew Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

Revert "package/x11r7/xlib_libXvMC: remove xlib_libXv dependency"

This reverts commit 52ac6d1c6bc625d8438539a7ff0f35d2057e9acc.

While looking at mesa3d package, I found this commit [1] about libxv
refering to an upstream commit [2].

The commit message is confusing "This library does not itself require
libXv for anything." but xv is still checked by the configure script
(pkg-config) and libXvMC still include Xvlib.h from XvMClib.h

In file included from XvMClibint.h:7:0,
                 from XvMC.c:5:
../include/X11/extensions/XvMClib.h:5:10: fatal error: X11/extensions/Xvlib.h:
No such file or directory
 #include <X11/extensions/Xvlib.h>
          ^~~~~~~~~~~~~~~~~~~~~~~~
compilation terminated.

So, xlib_libXv sill use xv library by including a header and using
XvImageFormatValues.

[1] https://cgit.freedesktop.org/mesa/mesa/commit/?id=2dbf10ba3d129d59e1a6c0e46b24c20629b277c6
[2] https://gitlab.freedesktop.org/xorg/lib/libxvmc/commit/dd9ae03725990ee9c9336bca4b267dfbb696002f

Fixes:
http://autobuild.buildroot.org/results/305/3051f4fbc1ab725de542a550e4f9eeb3cae385b8

Signed-off-by: Romain Naour <romain.naour@smile.fr>
Cc: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/flannel: remove 'v' prefix

Fixes version parsing for release-monitoring.org support.

Cc: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>
Cc: Heiko Thiery <heiko.thiery@kontron.com>
Signed-off-by: Matthew Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/pru-software-support: remove 'v' prefix

Fixes version parsing for release-monitoring.org support.

Cc: Heiko Thiery <heiko.thiery@kontron.com>
Signed-off-by: Matthew Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/dt: remove 'v' prefix

Fixes version parsing for release-monitoring.org support.

Signed-off-by: Matthew Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/gst1-plugins-bayer2rgb-neon: remove 'v' prefix

Fixes version parsing for release-monitoring.org support.

Signed-off-by: Matthew Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/perl-net-snmp: remove 'v' prefix

Fixes version parsing for release-monitoring.org support.

Signed-off-by: Matthew Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/spidev_test: remove 'v' prefix

Fixes version parsing for release-monitoring.org support.

Signed-off-by: Matthew Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/softether: remove 'v' prefix

Fixes version parsing for release-monitoring.org support.

Cc: Heiko Thiery <heiko.thiery@kontron.com>
Signed-off-by: Matthew Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/dav1d: bump version to 0.5.1

Changelog: https://code.videolan.org/videolan/dav1d/blob/master/NEWS

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/doxygen: bump version to 1.8.16

Added license hash, updated _SITE.

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/eudev: bump version to 3.2.9

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

utils/check-package: report := that appends to variables

Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>

fs: don't use := when not needed

Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>

package/gcc: don't override a variable

Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>

package/zmqpp: don't override variable

Instead of reseting the variable when we can't set it, just set it when
we can: we just have to negate the condition.

Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>

package/ligpgme: don't use :=, use += and subst

This makes it more in-line with how we handle such case in the
rest of Buildroot.

Also, move the variable assignment before its use. Even though
in a makefile that does not matter, it makes it cleaner, more
logical, and easier to follow (for us mere humans).

Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>

package/libidn2: get rid of remnants of using := in LICENSE

There is no longer any reason to assign to LICENSE using an
immediate assignment. There was even no reason to use it
either in the first assignment.

Drop it.

Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>

package/gstreamer1/gst1-plugins-bad: get rid of remnants of using := in LICENSE

There is no longer any reason to assign to LICENSE using an
immediate assignment. There was even no reason to use it
either in the first assignment.

Drop it.

Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>

package/*: do not override LICENSE, but append to it

Now that the infra takes care of removing extraneous spaces before
commas, we can safely append to LICENSE variables.

This removes the only case where we had to use immediate assignment
to append to a variable.

Mechanical change, with:

    $ sed -r -i -e 's/\<([^[:space:]]+_LICENSE) := \$\(\1\),/\1 += ,/' \
        $(git grep -l -E '\<([^[:space:]]+_LICENSE) := \$\(\1\),')

Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>

core/legal-info: don't require overriding _LICENSE

Currently, the formatting we impose on the _LICENSE variable requires
that we also use the rarely used := assignment operator, which makes
the _LICENSE variable the only variable that users have to write with
this operator.

This really departs from the simplicity and consistency of using the
append-assignment, which we use for every other variable.

This is because the append-assignment operator surreptiously
introduces a space between the original value and the appended one. But
we can use this knowledge, to match any instance of a space followed by
a comma, and turn it into a single comma.

This allows users to now have a consistent use of the '=' and '+='
operators we use everywhere else in .mk files.

Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>

package/libopenssl: set no-asm with generic architectures

Use no-asm when building with generic architectures such as gcc or
linux-generic32, see https://github.com/openssl/openssl/issues/9839

This will fix a static build failure on x86_64 due to the removal of
x86/x86_64 BSAES and AES_ASM support by
https://github.com/openssl/openssl/commit/87bea6550ae0dda7c40937cff2e86cc2b0b09491

Fixes:
 - http://autobuild.buildroot.org/results/e4f04bb13ec1b82b73db645bea4933e52bca4185

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/ccache: bump version to 3.7.5

Release notes:
https://ccache.dev/releasenotes.html#_ccache_3_7_5

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/leveldb: fix build for gcc < 5.x

Without defining LEVELDB_HAS_PORT_CONFIG_H port/port_config.h will not
be included:
https://github.com/google/leveldb/blob/master/port/port_stdcxx.h#L8

which breaks endian detection. Add the define to our memenv patch to fix
building with gcc < 5.
__has_include was added in gcc5: https://gcc.gnu.org/gcc-5/changes.html

Fixes:
http://autobuild.buildroot.net/results/4ac/4ac8fbed79307326086ab294eba9e68cf208e6ac/

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/libopenh264: fix mips build

Although this PR is still discussed upstream it

Fixes:
http://autobuild.buildroot.net/results/081/081c0426dde94015f5391e612454fc0b86888943/

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/glog: bump to version 0.4.0

Also fixed a build issue reported by test-pkg.

We now need to AUTORECONF as the pre-generated configure script is no
longer provided in the Git repository.

Signed-off-by: Gilles Talis <gilles.talis@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/gflags: bump to version 2.2.2

Signed-off-by: Gilles Talis <gilles.talis@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/check: bump version to 0.13.0

Release notes: https://github.com/libcheck/check/releases

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/ncdu: bump to version 1.14.1

Signed-off-by: Gilles Talis <gilles.talis@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/xapian: bump to version 1.4.13

Signed-off-by: Gilles Talis <gilles.talis@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/ccid: bump version to 1.4.31

Changelog: https://salsa.debian.org/rousseau/CCID/blob/master/README.md

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/linux-zigbee: use official tarball

- Use official tarball instead of using git
- Add linux-zigbee.hash

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

configs/stm32mp157_dk: bump U-Boot to 2019.10

Add patch to fix mounting rootfs partition properly.
This issue is already reported upstream.

Signed-off-by: Bartosz Bilas <b.bilas@grinn-global.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

board/stmicroelectronics/stm32mp157-dk: rename uboot partition in genimage.cfg

Change partition name according to the U-Boot requirements.

Signed-off-by: Bartosz Bilas <b.bilas@grinn-global.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

packages: host gcc >= 4.8 is now guaranteed

... so we can drop all config options about it and previous versions.

Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Cc: Arnout Vandecappelle <arnout@mind.be>
Cc: Peter Korsgaard <peter@korsgaard.com>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Romain Naour <romain.naour@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

docs/manual: bump our gcc/g++ requirements to 4.8

Currently, we only require a gcc 4.4 version, which now is pretty old
(released in April 2009).  This requirement is not even tested nowadays,
with our oldest autobuilder having a 4.7 version only.

And even then, 4.7 is still old enough that it prevents us from
upgrading some packages. For example cmake 3.10+ requires C++11
constructs that were only added in gcc 4.8 (when C++11 support was
finally completed in gcc).

So, update our requirements for gcc to at least 4.8.

Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Arnout Vandecappelle <arnout@mind.be>
Cc: Peter Korsgaard <peter@korsgaard.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/gvfs: fix static build with libarchive

Use pkg-config to retrieve static dependencies of libarchive in gvfs.mk
instead of patching configure.ac as upstream switched to meson

Fixes:
 - http://autobuild.buildroot.org/results/486187bb7b8800f44a8c173af518b84e68e53584

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/supervisor: fix license

trackrefs module has been removed since version 4.0.0 and
https://github.com/Supervisor/supervisor/commit/6c1c0adadaf190e77dc017a8a40c90249f0a055b

Since this removal, supervisor does not contain any ZPL-2.1 code
anymore, see
https://github.com/Supervisor/supervisor/commit/abff4cd872d2cbd1b999c1f439fdf83b4695cc6c

Fixes:
 - http://autobuild.buildroot.org/results/540303f884d9f891f58c660da22e2273956a92a6

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/rng-tools: powerpc64le requires libgcrypt

This patch adds the libgcrypt dependency for rngd_darn support
when building for the specific BR2_powerpc64le archtecture.
This entropy source is only enabled when powerpc64le is selected.

Fixes:
http://autobuild.buildroot.net/results/a41/a412ccef864e8279dcc917e106b5f7b515ed95da/

Signed-off-by: Matthew Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/mcelog: bump to version 165

- Switch site to github
- Use new LICENSE file added in version 160 with
  https://github.com/andikleen/mcelog/commit/f402998fa6943f3c4c4a2903d1ac85bcd92df75f

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/file: add upstream security fix

Fixes the following security vulnerability:

- CVE-2019-18218: cdf_read_property_info in cdf.c in file through 5.37 does
  not restrict the number of CDF_VECTOR elements, which allows a heap-based
  buffer overflow (4-byte out-of-bounds write).

Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/mjpegtools: fix build with -fPIC on x86

Retrieve upstream patch to fix build with BR2_PIC_PIE on x86

Fixes:
 - http://autobuild.buildroot.org/results/b1252b7907944d227694ad8506de25881bf2ca04

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

docs/manual: python is no longer a mandatory dependency

Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Cc: Arnout Vandecappelle <arnout@mind.be>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Peter Korsgaard <peter@korsgaard.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

support/dependencies: don't check for python on the host

We no longer have anything that needs it during the build, so we don't
require it anymore.

Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Cc: Arnout Vandecappelle <arnout@mind.be>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Peter Korsgaard <peter@korsgaard.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

core: drop check-uniq-files

Back a few years ago, when we were starting to think about top-level
parallel build, we were not sure how to deal with packages that
installed the same files, so we wanted to catch the situation to assess
how prevalent that was, before we decided what to do and how to address
it.

However, the trend nowadays is that packages will install in a
per-package target/ (and staging/ and host/), and the final directories
will be assembled in a reproducible (alphabetical) order, so if two
packages install the same file, the last one will win (as is currently
the case).

Besides, check-uniq-files reports loads of spurious errors when packages
get reinstalled (e.g. during development).

Finally, check-uniq-files is the only script called during the build,
that is written in python.

So, get rid of check-uniq-files.

Signed-off-by: Yann E. MORIN <yann.morin.1998@free.fr>
Cc: Arnout Vandecappelle <arnout@mind.be>
Cc: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Cc: Peter Korsgaard <peter@korsgaard.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/dstat: invert select of the python interpreter version

Commit 05484f621c1c2f80b2d735b2f4694996baad4a61 ("package/dstat: bump
to 0.7.4 which allows to run on Python3") added:

   select BR2_PACKAGE_PYTHON if !BR2_PACKAGE_PYTHON3

Unfortunately, some other packages do it the other way around, which
causes the following Kconfig circular dependency:

package/python/Config.in:5:error: recursive dependency detected!
package/python/Config.in:5: symbol BR2_PACKAGE_PYTHON is selected by BR2_PACKAGE_PYTHON3
package/python3/Config.in:5: symbol BR2_PACKAGE_PYTHON3 is selected by BR2_PACKAGE_PYTHON

Fix that by inverting the python version selection in the dstat
package.

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/dtc: fix license hashes

Update hash of license files and add BSD-2-Clause file, see
https://git.kernel.org/pub/scm/utils/dtc/dtc.git/commit/?id=25bb080c18d1a8a96760e26257bf8c80ec7fcab4
https://git.kernel.org/pub/scm/utils/dtc/dtc.git/commit/?id=702c1b6c0e73d2bcf24f0b8398aca6a940863e48

Fixes:
 - http://autobuild.buildroot.org/results/d3db8ddc096f09bb1728a0ba40b925fbf49c424b

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/busybox: bump version to 1.31.1

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

package/bison: bump version to 3.4.2

Release notes: https://savannah.gnu.org/forum/forum.php?forum_id=9554

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

configs/imx6-sabreauto: Add openssl for kernel build

Select BR2_LINUX_KERNEL_NEEDS_HOST_OPENSSL in order to fix
the following build error:

scripts/Makefile.host:90: recipe for target 'scripts/extract-cert' failed
make[2]: *** [scripts/extract-cert] Error 1
make[2]: *** Waiting for unfinished jobs....

Fixes: https://gitlab.com/buildroot.org/buildroot/-/jobs/332655822
Signed-off-by: Fabio Estevam <festevam@gmail.com>
Signed-off-by: Peter Korsgaard <peter@korsgaard.com>

DEVELOPERS: add Peter Korsgaard for docker-compose test

Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/docker-compose: bump version to 1.24.1

Fixes https://gitlab.com/buildroot.org/buildroot/-/jobs/332656041

The recent bump of a number of python packages broke docker-compose, as
docker-compose specifies both minimum and maximum versions for (most of) its
dependencies:

Dependencies of docker-compse 1.20.1 (! = unmet):
cached-property: < 2 (currently 1.51)
docopt: < 0.7 (currently 0.6.2)
! pyyaml: < 4.0, patched to < 4.3 (currently 5.1.2)
requests: < 2.19, patched to < 3 (currently 2.22.0)
! texttable: < 0.10 (currently 1.6.2)
websocket-client: < 1.0 (currently 0.56.0)
! docker: < 4.0 (currently 4.1.0)
dockerpty: < 0.5 (currently 0.4.1)
six: < 2 (currently 1.12.0)
jsonschema: < 3 (currently 2.5.1)
enum34: < 2 (currently 1.1.6)
backports.ssl-match-hostname: >= 3.5 (currently 3.7.0.1)
ipaddress: >= 1.0.16 (currently 1.0.23)

To fix this, bump docker-compose to the most recent release (1.24.1).  This
is unfortunately not enough, as our docker, pyyaml, requests and texttable
packages are too new, so add 3 patches from upstream to relax the version
checks of dependencies.  Notice that patch 0003 is from
https://github.com/docker/compose/pull/6623 and has not been merged yet.

Discussions around the problem of these maximum versions of the dependencies
and the fact that all downstream users have to patch it is ongoing here:

https://github.com/docker/compose/issues/6756

docker-compose 1.24.1 added a requirement for ssh support in python-docker in:
https://github.com/docker/compose/commit/7b82b2e8c721010b73f664e9d4657746a1fcd92b

So add a dependency for python-paramiko and update the toolchain dependency
for C++ (from python-paramiko -> python-cryptography) and adjust the
toolchain configuration of the runtime test to match.

Signed-off-by: Peter Korsgaard <peter@korsgaard.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/nodejs: bump to version 12.13.0

Signed-off-by: James Hilliard <james.hilliard1@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/Makefile.in: remove host-cc-option macro

https://git.buildroot.net/buildroot/commit/?id=91a08ecc998ae232ea6f3525540ed129d8176d18
added a macro solely used by the aespipe package:
https://git.buildroot.net/buildroot/commit/?id=00ecd72c28f103fc7d166f718db81a8b6c4919fa

After the latest version bump of aespipe the package does not need this
macro anymore so we can safely remove it.

Suggested-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>
Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/systemd: create /etc/resolv.conf only if resolved is selected

Or else it becomes a dangling link to
/run/systemd/resolve/resolv.conf, which is never created. Even worst,
it also prevents NetworkManager from updating resolv.conf.

Fixes:
  https://bugs.busybox.net/show_bug.cgi?id=9881

Signed-off-by: Carlos Santos <unixmania@gmail.com>
Reviewed-by: Yann E. MORIN <yann.morin.1998@free.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/meson: force ext static libs when BR2_STATIC_LIBS

This patch adds a case in the library pattern logic to prefer static
libraries when the Meson Core option for "default_library" is set to
solely static.

The existing library search order makes sense for cases of shared and
shared / static mixed. However if using a prebuilt cross-toolchain,
they usually provide both a static and shared version of sysroot
libraries. This presents a problem in a complete static build where
there won't be shared libraries at runtime and during build time there
are failures like "ld: attempted static link of dynamic object".

Bug:
https://github.com/mesonbuild/meson/issues/6108

Fixes:
http://autobuild.buildroot.net/results/db1740b4777f436324218c52bc7b08e5c21b667d/
http://autobuild.buildroot.net/results/c17/c17bbb12d9deadd64a441b36e324cfbbe8aba5be/

Cc: Arnout Vandecappelle <arnout@mind.be>
Signed-off-by: Matthew Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/dstat: bump to 0.7.4 which allows to run on Python3

Signed-off-by: Titouan Christophe <titouan.christophe@railnova.eu>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/supervisor: can now run on Python 3

Supervisor introduced support for Python3 starting with version 4.

Signed-off-by: Titouan Christophe <titouan.christophe@railnova.eu>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/supervisor: bump to v4.1.0

Signed-off-by: Titouan Christophe <titouan.christophe@railnova.eu>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

configs/friendlyarm_nanopi_neo_plus2: bump Linux/U-Boot

This updates Linux kernel from 5.0 to 5.3.4 and U-Boot from 2019.01 to
2019.10.

Signed-off-by: Louis Aussedat <aussedat.louis@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/flatbuffers: remove 'v' prefix from github-fetched packages

Fixes version parsing for release-monitoring.org support.

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

toolchain/toolchain-wrapper: explicitly pass --build-id=none if BR2_REPRODUCIBLE

Build ID is added to binaries at link time. Building in different
output directories causes some packages to have different Build IDs,
thus resulting in non-reproducibility.

Adding "-Wl,--build-id=none" fixes this issue by disabling setting of
Build ID.

Diffoscope output for Build ID issue:
https://gitlab.com/snippets/1886180/raw

After this patch, build is reproducible - i.e. diffoscope does not
produce any output.

Signed-off-by: Atharva Lele <itsatharva@gmail.com>
Reviewed-by: Romain Naour <romain.naour@smile.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

support/testing: add luvi test

Signed-off-by: Francois Perrad <francois.perrad@gadz.org>
Reviewed-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Tested-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

support/testing: add lua-sdl2 test

Signed-off-by: Francois Perrad <francois.perrad@gadz.org>
Reviewed-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Tested-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

support/testing: add lua-gd test

Signed-off-by: Francois Perrad <francois.perrad@gadz.org>
Reviewed-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Tested-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

support/testing: add lzlib test

Signed-off-by: Francois Perrad <francois.perrad@gadz.org>
Reviewed-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Tested-by: Ricardo Martincoski <ricardo.martincoski@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/uvw: new package

A header-only libuv wrapper in C++.

Signed-off-by: Asaf Kahlon <asafka7@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

utils/genrandconfig: use randint instead of (undefined) randInt

Fixes https://gitlab.com/buildroot.org/buildroot/-/jobs/333788455

Signed-off-by: Asaf Kahlon <asafka7@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/systemd: bump version to 243

Added config option for new systemd-pstore feature.

Removed patches that have been committed upstream.

Added symlinks required to enable some services.

After extensive discussion with upstream it does not seem there is any
acceptable option to have symlink installation handled on their side.

The recommended solution from upstream is to have systemctl handle
service installation, however this has a number of downsides such as
requiring us to build a host-systemd just to install a few symlinks.

Since we already customize service installation it is simpler for us
to create the symlinks ourselves, this will also simplify service
customization on our side as we will not need to disable any systemd
default services in cases where we need to make customizations.

In addition using systemctl would introduce a minimum host headers
dependency in which we don't have the infrastructure to handle
properly.

The downside to this approach is that we need to track service creation
changes from upstream ourselves, however this is relatively straightforward
and since upstream service file locations are relatively stable regressions
should be limited. This will mostly be a concern in regards to adding
support for new systemd features.

Details:
https://github.com/systemd/systemd/issues/12767
https://github.com/systemd/systemd/pull/12164
https://github.com/systemd/systemd/pull/12769
https://github.com/systemd/systemd/pull/12775

Signed-off-by: James Hilliard <james.hilliard1@gmail.com>
Tested-by: Jérémy Rosen <jeremy.rosen@smile.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/python-libconfig: fix build with python 3.8

Fixes:
 - http://autobuild.buildroot.org/results/edf32c178b7912a987e119f776a51f464424dceb

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/dtc: bump to v1.5.1

Remove 0002-Makefile-Add-EXTRA_CFLAGS-variable.patch which is now
applied upstream

Signed-off-by: Titouan Christophe <titouan.christophe@railnova.eu>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/scons: bump to version 3.1.0

Signed-off-by: Asaf Kahlon <asafka7@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/python-uvloop: force re-cythonize on build

The generated loop.c provided with the tarball was generated with
Cython 0.29.10 and it doesn't compile against the new Python 3.8
headers. Upstream is now working on a new uvloop release which will
use the newest Cython (0.29.13).  Until then, we will remove the
current loop.c and force regenration of it.

Fixes:
 - http://autobuild.buildroot.net/results/bffe1b2ee7b674f870e15398680ce5b5b22fd296

Signed-off-by: Asaf Kahlon <asafka7@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/python-ipython: bump to version 7.9.0

And fix indentation in hash file.

Signed-off-by: Asaf Kahlon <asafka7@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/python-docutils: add upstream patch to build without setuptools

An upstream patch to enable building without setuptools is added.
The original commit is adjusted slightly to fit the current version.

Fixes:
 - http://autobuild.buildroot.net/results/8f3c01f0e286c6b15c8c973f91dd69ee816a58b9

Signed-off-by: Asaf Kahlon <asafka7@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/arp-scan: bump version to 1.9.6

Removed patch which was applied upstream:
https://github.com/royhills/arp-scan/commit/cbeda6a9c5e154fd5e5d56e3e6e6bdc5ddd2b088

We still need autoreconf however due to using source code retrieved by
the github helper.

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/aespipe: bump version to 2.4f

Updated link to package homepage since old link is just a redirect now.

Added sha1 and md5 hashes provided by upstream.

Removed -no-pie workaround after upstream added a check in configure,
quoting ChangeLog:

- Added configure script autodetection for x86/amd64 -fno-pie compile
  and -no-pie linker flags. These are needed when compiler/linker
  produces position independent executable (PIE) by default.

Build-tested host-aespipe on Debian Buster 10.1 using
"gcc version 8.3.0 (Debian 8.3.0-6)"

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/acsccid: bump version to 1.1.7

Added hashes for tarball and license file.

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/acpica: bump version to 20191018

Rebased patch due to upstream commit
https://github.com/acpica/acpica/commit/36037bdaf0ab843b7304d538d4e5cbc0c9524024

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

utils/genrandconfig: test configurations with BR2_PIC_PIE

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/librsvg: fix build with NLS

Fixes:
 - http://autobuild.buildroot.org/results/9fb04aa7761845a58df14d460b4a7bbf426e19ee

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/f2fs-tools: bump to 1.13.0

This release includes:
 - enable VERITY by default on Android
 - introduce some preen mode in fsck.f2fs
 - add f2fs_io tool
 - add casefolding support

And, there are lots of bug fixes.

Signed-off-by: Grzegorz Blach <grzegorz@blach.pl>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/linux-firmware: embed all RS9113 firmware files to target image

Only rs9113_wlan_qspi.rps firmware was present on generated image.

Signed-off-by: Adrien Ricciardi <adrien.ricciardi@hotmail.fr>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/brltty: fix build with expat

tbl2hex is a host command line that is built with:

$(CC_FOR_BUILD) $(LDFLAGS_FOR_BUILD) -o $@ $(TBL2HEX_OBJECTS) $(EXPAT_LIBS_FOR_BUILD) $(LDLIBS_FOR_BUILD)

It needs cldr.o which can depends on expat and is built through:

%.$B: $(SRC_DIR)/%.c
       $(CC_FOR_BUILD) -DFOR_BUILD $(CFLAGS_FOR_BUILD) -o $@ -c $<

When cross-compiling, build fails because expat is not found on host:

gcc -DFOR_BUILD -I. -I. -I./../Programs -I../Programs -I../Headers -I./.. -I.. -D_LARGEFILE_SOURCE -D_LARGEFILE64_SOURCE -D_FILE_OFFSET_BITS=64  -D_DEFAULT_SOURCE -D_POSIX_C_SOURCE=2 -D_BSD_SOURCE -D_XOPEN_SOURCE=500 -D_XOPEN_SOURCE_EXTENDED -D_GNU_SOURCE  -D_DEFAULT_SOURCE -D_POSIX_C_SOURCE=2 -D_BSD_SOURCE -D_XOPEN_SOURCE=500 -D_XOPEN_SOURCE_EXTENDED -D_GNU_SOURCE -DHAVE_CONFIG_H -g -O2 -std=gnu99 -Wall -Werror=format-security -o cldr.build.o -c cldr.c
cldr.c:31:10: fatal error: expat.h: No such file or directory
 #include <expat.h>
          ^~~~~~~~~

To fix this issue, add host-expat dependency as well as patch so that
the *_FOR_BUILD variables are properly passed to the sub-configure
script that brltty runs to build the host utilities.

Fixes:
 - http://autobuild.buildroot.org/results/362cfb57e4a91a066493269d8078d931529ddf69

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
[Thomas: use a different, simpler solution.]
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/kmod: needs AUTORECONF = YES

Commit 95fd2b593ef7194a23a20a64dac1e21e4e650ef2 ("package/kmod: add
patch fixing the build with Python 3.8") added a patch that touches
Makefile.am, but forgot to add AUTORECONF = YES. Let's fix this.

Fixes:

  http://autobuild.buildroot.net/results/5185deba0c18072928a0a6867f15f77448228b0b/

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/bird: fix build without babel, bgp, ospf, radv, rip or static

Fixes:
 - http://autobuild.buildroot.org/results/0b00948eed9bb8405b70f3f9112ecce99b365f35

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/libselinux: fix the build with Python 3.8

Following the switch to Python 3.8, the libselinux Python extension
started to fail building. This is fixed by upstream commit
2efa06857575e4118e91ca250b6b92da68b130d5, which we backport as
0003-libselinux-Use-Python-distutils-to-install-SELinux-p.patch.

This patch has the nice merit of switching to using distutils to build
the Python extension of libselinux, instead of some custom logic. This
allows to significantly simplify our libselinux.mk: we can rely on
PKG_PYTHON_DISTUTILS_ENV and HOST_PKG_PYTHON_DISTUTILS_ENV instead of
lots of custom variables.

However, upstream commit 2efa06857575e4118e91ca250b6b92da68b130d5 had
its own issues:

 * Hardcode of -I $(DESTDIR)/$(INCLUDEDIR) -L $(DESTDIR)/$(LIBDIR) at
   build time, while DESTDIR is normally empty at build time, causing
   bogus -I /usr/include -L /usr/lib to be used

   This is fixed in
   0004-src-Makefile-don-t-pass-bogus-I-and-L-to-python-setu.patch

 * New usage of ln --relative, which is not supported in older
   distributions.

   This is fixed in
   0005-Remove-ln-relative-usage-in-install-pywrap.patch

 * Usage of the host Python "imp" module to query the extension used
   for native Python module, but that returns an incorrect result when
   cross-compiling. We chose to simplify the code to not have to query
   for this information.

   This is fixed in
   0006-Do-not-use-PYCEXT-and-rely-on-the-installed-file-nam.patch

With this patch, the libselinux Python module was built-tested with
Python 2 and Python 3, and run-time tested as well in both
configurations, for both the target and host variants of libselinux.

Fixes:

  http://autobuild.buildroot.net/results/aeb58de7ad674b980258e6ed30c7da3949a04452/

Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/mesa3d: remove upstream patch

Remove 0004-meson-Link-xvmc-with-libxv.patch witch was backported to mesa3d
19.2. This patch was added to Buildroot at the time when mesa3d version 19.1
was used.

Signed-off-by: Romain Naour <romain.naour@smile.fr>
Cc: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/micropython: require python3 on the host

There are some scripts used when building micropython that require
python3 on the build host. Use BR2_PYTHON3_HOST_DEPENDENCY so this can
be either be satisfied by installing it on the build host or by building
the host-python3 package.

Fixes:
- http://autobuild.buildroot.net/results/b85b2214576030026a8e04d142a77a2648379417/

Signed-off-by: Chris Packham <judge.packham@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/znc: set WANT_SWIG

Set WANT_SWIG={ON,OFF} to {en,dis}able swig and avoid a random build
failures probably due to parallel build issue when extracting
pre-generated tarball:

CMake Error: Problem with archive_write_header(): Can't unlink already-existing object
CMake Error: Current file: swigpyrun.h
CMake Error: Problem extracting tar: /usr/lfs/hdd_v1/rc-buildroot-test/scripts/instance-0/output/build/znc-1.7.5/modules/modpython/generated.tar.gz

This tarball contains pre-generated files, and is not used when
host-swig is available.

Fixes:
 - http://autobuild.buildroot.org/results/f3394de616cea4f474b6d5887aa0d5f5dff05281

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/qt5/qt5base: fix icu detection in Qt 5.6.3

icu is required to build Q5tWebkit.
When UChar is defined as char16_t in ICU, then qtbase fails to detect ICU.
The issue is described https://bugreports.qt.io/browse/QTBUG-49586

Build fails with following error messages:
...
ustring.h:473:20: error: ‘UChar’ does not name a type
 u_strCompare(const UChar *s1, int32_t length1,
                    ^
          ^
make[2]: *** [Makefile:195: icu.o] Error 1
ICU disabled.
The ICU library support cannot be enabled.

Signed-off-by: Andrey Skvortsov <andrej.skvortzov@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/subversion: bump version to 1.12.2

Added dependencies to lz4 and utf8proc.

Replaced the 0002-disable-macos-specific-features.patch by a simpler
patch/workaround that still works after the version bump.

Updated license hash after various upstream commits:
https://github.com/apache/subversion/commits/trunk/LICENSE

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/kvm-unit-tests: fix build on Arch Linux x86_64

On x86_64, we use the host compiler instead of the target compiler to
build kvm-unit-tests, because it is built with -m32 and our target
compiler doesn't support that.

However, the compiler on Arch Linux is broken: it *always* builds with
-fstack-protector, even when -ffreestanding is passed. However, when
-fnostdlib is passed at link time (which is normally the case when
building with -ffreestanding), it is not linked with the stack-protector
library. This leads to a link time error:

/usr/bin/ld: x86/realmode.o: in function `print_serial_u32':
.../x86/realmode.c:104: undefined reference to `__stack_chk_fail'

Since the entire package is built with -ffreestanding, it doesn't
support stack-protector at all. Therefore, simply pass
-fno-stack-protector explicitly on x86_64 to work around the bug in Arch
Linux.

Bug reported upstream: https://bugs.archlinux.org/task/64270

Fixes:
 - http://autobuild.buildroot.org/results/e6f767755ffdb5ecc014eb5ad7519814f075a60e

Signed-off-by: Fabrice Fontaine <fontaine.fabrice@gmail.com>
Tested-by: Matthew Weber <matthew.weber@rockwellcollins.com>
Signed-off-by: Arnout Vandecappelle (Essensium/Mind) <arnout@mind.be>

package/leveldb: bump version to 1.22

Upstream switched to cmake, removed patches for the old buildsystem
and added new patch to install libmemenv.a and memenv.h.

Added license hash.

Package requires gcc >= 4.8:
https://github.com/google/leveldb/blob/master/CMakeLists.txt#L14

Removed "v" from LEVELDB_SITE to reflect current naming scheme.

Signed-off-by: Bernd Kuhls <bernd.kuhls@t-online.de>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>

package/acpid: refactor init script

Adapt the format to the current template, used in other init scripts.

Move the one socond delay in restart to stop, giving acpid time to send
dying gasp to syslog.

Users willing to add start arguments can set the ACPID_ARGS variable in
/etc/default/acpid instead of rewriting the init script.

Signed-off-by: Carlos Santos <unixmania@gmail.com>
Signed-off-by: Thomas Petazzoni <thomas.petazzoni@bootlin.com>